We propose to explore novel scintillation light detection concepts and develop a high performance imaging detector module suitable for a positron emission tomography (PET) system designed to image small laboratory animals. If successful this innovative detector will enable us to push the performance limits of high resolution PET. Recent studies suggest PET is a useful tool for molecular imaging of small laboratory animal models for human disease. Limitations of currently available PET systems for molecular imaging are non-optimal sensitivity and spatial resolution for identifying molecular signals of interest from minute structures in a reasonable time. Higher sensitivity may be achieved by bringing detectors closer and covering a larger fraction of the animal. Spatial resolution may be improved by using finer scintillation crystal array elements. Currently there are several groups developing PET scintillation arrays with 1 mm crystal pixels. The challenge with going to finer crystals is extracting sufficient light to maintain adequate detection signal-to-noise ratio (SNR). If the crystals are kept long to maintain high detection efficiency, using finer crystals results in significantly lower light collection efficiency and SNR. As long as the resulting light signals are sufficiently above noise, the finer crystals may still be resolved and the desired intrinsic resolution may be achieved. However, high spatial resolution alone is not sufficient. Weak light signals mean non-optimal sensitivity since more pulses are now below the detection threshold, and non-optimal energy and time resolutions, which may result in significant image contrast degradation. We propose a novel detector concept for PET that will facilitate ultra-high spatial resolution (< 1 mm), high coincidence count sensitivity (>14%), and optimal light collection efficiency (>95%) all at once. The design uses scintillation crystals coupled in an innovative manner to novel semiconductor photodetector arrays, and custom readout electronics. The new scintillation detection concepts will be investigated during the R21 phase. During the R33 phase, two high performance imaging detectors will be built and studied. If successful the proposed design will significantly improve efficiency of converting scintillation light into electrical signals. Together with the high sensitivity and spatial resolution proposed, the implications for its use in high resolution PET are substantial.